Idea Transcript
1
Fischer-Tropsch Synthesis Over Cobalt Catalysts: State of the Art and Examples from Natural Gas and Biomass Conversion to Fuels Edd A. Blekkan Dep. of Chemical Engineering, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, Norway Metal Kokkola 2013
2
Fischer-Tropsch chemistry: nCO+(2n+2)H2 •
•
Co, Fe, Ru, Ni
-(CH2)n- + nH2O
Catalytic, reductive polymerization of C1 units • 1 < n < ~ 50 – 100 • Hydrocarbon products: N-alkanes and linear 1-alkenes • Chain length governed by chain growth probability α Exothermic reaction (~ 165 kJ/mol CO converted)
3
Outline • Introduction – Fischer, Tropsch and a couple of other old men – Basic chemistry and technology
• Examples - mainly from our work – FTS experimental work – how to do – Effect of water – Catalyst parameters • Composition and pore size • Cobalt particle size
– BTL-related issues • Sensitivity to alkali • Scale issues - FTS in microstructured reactors
• Summary
4
FTS: History – discovery phase •
Sabatier 1902 –
•
Mittasch –
•
Mittasch & Schneider (BASF) patents 1913-1916
Fischer & Tropsch – – –
•
P. Sabatier and J.B. Senderens, Compte Rendus Hebd., 134 (1902) 514. (Methanation)
F. Fischer and H. Tropsch, Brennst. Chem., 4 (1923) 193-197 (methanation). F. Fischer and H. Tropsch, Brennst. Chem., 4 (1923) 276-285 (Synthol). German Patent 484337 (1925) (Hydrocarbons)
The work of Fischer and Tropsch led to the industrial utilization of the process
Photos: From left: Fischer, Tropsch, Mittasch and Sabatier
5
Paul Sabatier Nobel price 1912 •
Quote from his Nobel lecture (December 1912): …During the period 1901 to 1905, together with Senderens, I showed that nickel is very suitable for the direct hydrogenation of nitriles into amines and, no less important, of aldehydes and acetones into corresponding alcohols. Carbon monoxide and carbon dioxide are both changed immediately into methane, which can therefore be synthesized with the greatest ease…
6
Most important recent developments
ORYX (top) (37 kbbl/day) and PEARL (140 kbbl/day) projects (Qatar, in production)
7
Synthesis Gas Feedstocks and Products Coal
H2 (or NH3) CH4
Gas
Synthesis Gas (CO+H2)
CH3(CH2)nCH3
Oil CH3OH Biomass CH3(CH2)nCH2OH
8
•
Chain growth polymerization mechanism explains the typical product distribution – Stepwise addition of single carbon atom units – Chain length determined by a single parameter α; the probability of chain growth
(1 − α ) 2 S ln = n ln ( α ) + ln α n – α varies with catalyst and conditions (T, H2:CO ratio etc. .)
• Catalysts – Ru • Highest activity, selectivity, low WGS • Not available for commercial use
– Fe • ”Cheap”, reasonable activity, but usually operated at ”high” T • Oxygen out as CO2 (WGS close to equilibrium)
– Co • High activity, good selectivity • Low T operation • Very low WGS activity – oxygen out as H 2O
– Ni • Methanation • Chain growth possible (low T)
• Modern Co catalysts – Supported on inorganic support • alumina-, silica- or titania-based
– 220-230 °C, 20-30 bar
9
Gas to liquids • Modern GTL based on low temperature FTS over cobalt-based catalysts – Maximizing wax (or liquid products), minimizing methane and light HCs – Refining (hydrocracking) wax to products
• Fixed bed or slurry reactors – Both have advantages and problem issues
• Syngas production: ATR or gasification
Air
Oxygen production
Natural gas/ steam
Synthesis gas production
H2 CO CO2
Raw products Fischer-Tropsch synthesis
Separation/ upgrading
(CO+2H2)n → -(CH2)n- + nH2O
Final products
10
FTS Reactors
11
CO activation, monomer form •
Key point of the mechanism – Understanding how the catalyst works can allow improvements
•
3 main theories (simplified description) – Carbene mechanism (*CO → *C + *O) • Carbon species can be CHX where X = 1,2 or 3 • Developed from the original carbide proposal by Fischer and Tropsch
– Enol mechanism (*CO + 2H* → *CHOH + 2*) • CO is adsorbed and hydrogenated to ”enol” and water is formed by condensation of adjacent species • Proposed by Anderson and Emmett
– CO insertion (*CO + H* → *COH +*) • CO is inserted into metal-methyl (methylene) and hydrogenated to form alcohol or alkene • Proposed by Pichler and Schulz
• •
Support for all theories in experimental work Ongoing theoretical efforts (DFT) – Conflicting results
• State of the art: We don’t know – still an open question
12
Some examples from our work
13
FTS Experimental work • Approx. 2 g catalyst in tubular micro-reactor, diluted with inert (SiC) • Pre-reduction for 16h at 350 °C, 1 bar H2 • 210°C, 20 bar pressure, H2/CO = 2.1 • Controlled start-up of the experiment (to avoid temperature run-away) • Fixed flow-rate for 24h (activity data) • Adjust feed rate to obtain 50% (or 40%) CO conversion (Selectivity data)
14
Effect of CO conversion on selectivity
Adapted from Storsæter, Borg, Blekkan, Holmen, J. Catal. 231 (2005), 405-419.
Valid comparison of selectivity can only be done at fixed CO conversion
15
Main reason: C5+ Selectivity increases with PH2O • Example: Co/CNF – 20%Co/CNF (inset TEM image) – IW impregnation of purified platelet CNF – Sg= 117 m2/g – DCo= 5.4% • (18 nm Co particles)
– FTS at 483 K, 20 bar, H2:CO = 2:1.
Borg, Yu, Chen, Blekkan, Rytter, Holmen, Topics in Catalysis, in press.
16
Effect of water •
Water is the main product of the FTS over Co – The actual partial pressure depends on reactor type and flow pattern
• •
Water adsorbs on the catalyst surface Water influences rate and selectivity – Higher PH2O always leads to more WGS – C5+ selectivity always improved with higher PH2O, to some extent due to reduced rate of methane formation – Most systems give enhanced overall rate with higher PH2O , but there are some reports of no effect or negative effect • Some indications of importance of pore size
•
Water increases deactivation – Due to: Sintering, restructuring, formation of nonreducible cobalt-support compounds (aluminate/silicate..) or oxidation?
Review papers: Water effect: Blekkan et al., Catalysis (RSC London) 20 (2007) 13. Water effect: Dalai and Davis, Applied Catalysis A: General 348 (2008) 1. Deactivation: Tsakoumis et al., Catalysis Today, 154, (2010) 162.
17
Effect of water on rate Formal kinetics •
Conventional LHHW kinetic equations are usually established without water term, eg. a recent paper from Botes (Sasol), for a 20Co-0,5Pt/Al2O3, 5 months into its life in a slurry reactor
•
A proposal from Lyon/Lille (Anfray et al, 2007) includes a water-term in the denominator (originally proposed by van Steen and Schulz, 1999)
– –
Water term ”not determined unambiguously” since water was not added, only the indigenous water pressure In this form the liquid phase concentrations (CiL) are used; established through gas-liquid equilibrium calculations Botes, Ind. Eng. Chem. Res. 48 (2009) 1859. Anfray et al. Chemical Engineering Science 62 (2007) 5353. van Steen, Schulz, Applied Catalysis A: General 186 (1999), 309.
18
Previous work (20 bar, H2:CO = 2.1)
• •
On γ-Al2O3 (left) we observe an apparent negative effect of 20% added water on rate On TiO2 (right) the effect is clearly positive –
•
SiO2 gives similar response (not shown)
Several others report similar findings (see e.g. Dalai and Davis) Storsæter et al., J. Catal. 231 (2005) 405. Dalai and Davis, Applied Catalysis A: General 348 (2008) 1.
19
More careful study – positive kinetic effect also on γ-Al2O3 (12Co-0.5Re/γ-Al2O3)
Dotted line: Added water Solid line: Total average water
• • •
Key difference from previous work: Inlet syngas pressure kept constant (18 bar) For each incremental increase in PH2O there is an increase in rate Deactivation is stronger in periods with added water S. Lögdberg et al., Applied Catalysis A: General 393 (2011) 109–121
20
Individual rates C1
C4 C3
• •
C2
Methane formation inhibited by water Total rate enhancement due to chain growth S. Lögdberg et al., Journal of Catalysis 274 (2010) 84–98
21
Water effect - summary •
Observations – A positive effect of addition of moderate amounts of water with all investigated systems (TiO2, SiO2, Al2O3 supports, with/without Re promoter, even with unsupported Co) – Kinetic effect: Enhanced overall rate – Selectivity improved because of enhanced chain growth and reduced methane formation – Deactivation and pressure effects may have masked the true kinetic effects in previous experiments with alumina support
•
The effect is common for all cobalt catalysts – Some influence of support, but key elements of the mechanism must be linked with the behaviour of the cobalt surface
• Mechanism: Still open to debate
22
Effect of catalyst parameters on selectivity • Alumina –supported 20% Co catalysts • 0% Re (filled symbols) or 0,5% Re (open symbols • Small, but clear increase in chain growth probability with Re addition – Slightly lower methane and C2C4 selectivity with Re
• Figure also illustrates effect of pore size of aluminasupported catalysts – Wide pore supports give higher SC5+
Ø. Borg et al., Catalysis Today 142 (2009) 70.
23
•
•
Incipient wetness impregnation preparation route leads to Co particle size strongly correlated with support pore size Unclear from this work if it is the particle size or pore size that influences selectivity
Borg et al., Catal. Today 142 (2009) 70. Idem, Catal. Lett. 126 (2008) 224.
24
Pore size effect confirmed Pore size (Å)
GHSV (NL/g,h)
CO conv. %
CH4 sel. %
C5+ sel. %
140
2,6
44.0
8.4
81.8
260
3.5
43.6
8.3
84.0
•
Co catalysts with similar particle size supported on mesoporous alumina – Alumina synthesized from isopropoxide using a structure directing agent – Material not ordered
•
Significant effect of pore size on SC5+
Lesaint et al. Applied Catalysis A: General 351 (2008) 131
25
Ru time yield 104 s-1
Co time yield 104 s-1
Particle size also matters? ”Normal” cobalt catalyst: TOF is constant
E. Iglesia, S.L. Soled, R.A. Fiato. J. Catal., 137 (1992) 212
26
Steady-State Isotopic Transient Analysis r
CO
μmol/g,s
¥
t i = ò Fi ( t ) dt 0
483 K, 1.85 bar, H2/CO/inert=15/1.5/33.5 0.5%Re
CH(4)
s
θ
CH(4
)
TOF
CH4
10-3s-1
1,0
12Co/γAl2O3
1.6
10
0.07
12CoRe/γAl 11 2O3 2.7 0,6 20CoRe/γAl2O3 4.1 9 0,4 20CoRe/αAl 12 2O3 1.6
0.10 0.11 0.12
7.1
0,8
F ( t )
TOF = k ×q= t - 1 ×q
τ
8.6 11.9 10.2
Ar Kr 12 CO 13 CO
0,2
12Co/SiO2 0,0 12CoRe/SiO2 0
10
1.0 1.3
13 0.10 11 0.10 20
30
Time [s] V.Frøseth, S. Storsæter, Ø.Borg, E.A. Blekkan, M. Rønning, A. Holmen, App.Catal. A: General 289 (2005) 10
7.5 8.7 40
27
Smaller particles behave differently • • •
•
Grey points from Iglesia et al. 1992.
De Jong and coworkers in Utrecht showed systematic change in behaviour below 6-8 nm, smaller particles give – –
•
Co/CNF 35 bar, H2:CO = 2
Lower specific activity Poorer C5+ selectivity, especially due to higher methane selectivity
What is the origin of this behaviour? –
Non-classical particle size effect
Bezemer et al. J. Am. Chem. Soc., 128 (2006), 3956. (The idea presented at NGCS VII in Dalian 2004; Bezemer et al. Stud. Surf. Sci. Catal. 147 (2004), 259.)
28
Small particles have different surface coverages
SSITKA demonstrates that small particles have – Higher coverage of ”irreversible” CO (blocks the surface, leads to lower activity) – Higher coverage of H (leads to more methane)
den Breejen et al, JACS, 131 (2009), 7197.
29
Proposed mechanism
• Cobalt modelled as cubo-octahedral particle • Smaller particles have larger fraction CUS atoms • Also affects terrace sites den Breejen et al, J. Am. Chem. Soc., 131 (2009), 7197.
30
Thermodynamic explanation? •
•
•
Particles change character when the size is reduced below about 100 atoms – this is reflected in the chemical potential (surface energy and Laplace pressure) The mechanism is reduced to kinetically significant steps, and modelled in terms of changes due to adsorption
The graph shows the data (from Bezemer et al.) and a fit to two models assuming two significant kinetic steps (blue solid line) or a LHHW model (red dotted line). –
P1, P2 are lumped kinetic parameters, α is the power in the Brønsted equation k = gK α linking kinetics with thermodynamics and η is a material term containing surface tension and molar volume D.Yu. Murzin, Chemical Engineering Science 64 (2009) 1046. Parmon, Dokl.Phys.Chem. 413 (2007),42.
31
Hypothesis • The «non-classical» particle size effect appears to reflect differences in adsorption properties • We should be able to relate the heats of H2 and CO adsorption with the catalytic properties and Co particle sizes?
Patanou, Tveten, De Chen, Holmen, Blekkan; Catalysis Today, 214, (2013), 19.
32
Catalysts
One-step incipient wetness impregnation 20%wt. Co on γ-Al2O3 with different metal particle sizes
H2O/EG
XRDb
DOR
Dispersion
Corrected PSc
Catalysts
mass ratio
(nm)
(%)
(%)
(nm)
Co-5
0,80
4,2
45
9,0
4,8
Co-6a
0,91
6,5
52
7,8
6,4
Co-10
0,93
10,8
59
5,7
9,9
Co-13
0,96
12,2
73
5,6
12,6
Co-17a
0,96
17,0
81
4,7
16,7
large pore alumina (Puralox TH high pore volume series) Co particle size diameter calculated by the theoretical shrinking of the crystallite d(Co0)[nm] = 0.8 · d(Co3O4) c corrected Co particle size diameter for the degree of reduction d(Co 0)[nm] = 96.2 · (DOR/%D) a b
Patanou, Tveten, De Chen, Holmen, Blekkan; Catalysis Today, 214 (2013), 19.
33
H2 adsorption at 313Κ - Co particle size effect
Patanou, Tveten, De Chen, Holmen, Blekkan; Catalysis Today, 214 (2013), 19.
34
H2 adsorption at 313Κ – ΔHads as function of coverage Two district regimes Low coverage regime below 10% Initial ΔHads of H2 118 - 139±3 kJ/mol “Linear” decline in ΔHads with increasing coverage beyond 10% But: No significant difference between the particle sizes
Patanou, Tveten, De Chen, Holmen, Blekkan; Catalysis Today, 214 (2013), 19.
35
CO adsorption at 313Κ - Co particle size effect
Patanou, Tveten, De Chen, Holmen, Blekkan; Catalysis Today, 214 (2013), 19.
36
CO adsorption at 313Κ – heats of adsorption
Two district regimes Low coverage regime up to 5% Initial ΔHads of CO 121 - 141±3 kJ/mol Almost constant ΔHads in the range 5% to 80% coverage (90 – 100 kJ/Mol) But: No significant difference between cobalt particle sizes
37
Cobalt particle size effect on catalyst performance
FTS conditions: 210 °C, 20 bar, H2/CO=2, fixed bed reactor, selectivities compared at 40% CO conversion
small differences in the activity (TOF) results
marked C5+ selectivity variations
weak effect of particle size on activity, increase in C5+-selectivity with increasing particle size
38
BTL
Figure: Kavalov B, Peteves SD, 2005. (EU report, ISBN 92-894-9784-X )
•
BTL introduces other issues – Economy of scale – Biomass contains a range of elements, cleaning/scrubbing at low temperature has a cost • Investment • Efficiency
•
Recent review: Lillebø, Holmen, Enger, Blekkan, WIREs Energy and Environment, Vol. 2 (2013), 507. (DOI:10.1002/wene.69)
39
Effect of alkali • Alkali species are important in syngas from biomass and coal gasification – Concentration and speciation depends on conditions and composition (Norheim et al.) – In case of poor design or operation of gas cleaning units alkali species can reach catalytic reactors
• Sodium can be present in alumina support materials, and has been shown to influence specific activity strongly (Borg et al.). A. Norheim et al. /Energy & Fuels 23 (2009) 920–925 Ø. Borg et al. / Journal of Catalysis 279 (2011) 163–173
40
Catalytic activity
• Even low concentrations lead to large loss in activity
-Lillebø, Patanou, Yang, Blekkan, Holmen, Catalysis Today, 215 (2013), 60. -Patanou, Lillebø, Yang, Chen, Holmen, Blekkan, Industrial & Engineering Chemistry Research, in press , dx.doi.org/10.1021/ie402465z
41
Selectivities (@ 50% CO conversion)
• The selectivities are also influenced by the additions – Increased chain growth, less CH4 – Increased CO2 (but still very low) – Alkali species similar, Ca has only low influence on HC distribution
42
Heats of adsorption: H2 adsorption at 313Κ - Sodium addition effect
-
Volumetric chemisorption results confirmed Initial ΔHads of H2 106 116 kJ/mol No significant difference between Na levels from 0 – 1000 ppm
43
Heats of adsorption: CO adsorption at 313Κ - Sodium addition effect
-
-
Initial ΔHads of CO 130 - 143 kJ/mol No significant difference between Na levels from 0 – 1000 ppm Adsorption ratio H2 : CO ~ 1 - Bridge bonding of CO
44
Effect of alkali - mechanism? • The addition of alkali /alkaline earth salts up to 1000 ppm modifies the catalytic activity and selectivity • Volumetric chemisorption and microcalorimetry do not show any differences within the uncertainty of the measurements • The additions represent a very low fraction of the number of surface sites – Except Li, which has a similar effect as the other additions on a weight basis
45
Site blocking?
A.G. Norris et al. / Surface Science 424 (1999) 74–81
• Hypothesis: – Corrugated and stepped sites are of particular importance for activity (Shetty et al., Catal. Today 171 (2011) 168.) – Small amounts of alkali have been shown to concentrate on steps (Norris et al., image on the left shows K on a Ni surface) – But: At surface science conditions metallic K is added, alkali salts probably leave alkali as ionic (+1) species • Blocking effect similar?
46
Microstructured reactors in FTS? • Microstructured reactors: – Large number of small, parallel channels • enhanced mass transfer properties • intensified heat transfer
– Different production technologies for these systems – Compact and modular • Suggested for offshore use, BTL, biogas conversion etc. • New paradigm in scaling?
• Suitable for FTS? – Catalyst introduction • Wall coating • Packed bed • Structured systems
– Pressure drop and stability issues?
• Companies claim good results – Oxford catalyst (Velocys) recipient of 2010 XTL award for their system
47
Oxford catalyst/SGC (Portugal) Demo in Güssing, Austria: • •
>900 microchannels ”Very high productivity”: 0.75 kg FT liquids per litre of catalyst and hour – Claim: 4-8 times greater productivity compared to conventional systems??? – Unit ”demonstrating robust responsiveness to shutdowns and start ups…”
http://www.engineerlive.com/Process-Engineer/Environmental_Solutions/Biofuel_processing_takes_a_step_closer_to_reality/23103/
48
Micro-channelled reactor Steel microstructured reactor (KIT, Karlsruhe) – 8 sections with 800 μm channel height – Channels formed by stacking of foils etched with a pillar structure – Cross-flow oil-channels between catalyst layers (cooling) – Unit (2 cm3) placed in tube fitting – Filled with catalyst (high loading Co/Al2O3, particle size 53-75 μm – Compared with same catalyst in FBR (diluted 1:20 with SiC to ensure uniform temperature, Δ